cluj

Now using the ESPN API, which is completely undocumented, but perhaps a better way to pull than data than web scraping.

Looks like PJ Washington's name might have changed from P.J. to PJ in either ESPN or bball reference. Add something to remove periods from names?

Player Rater:

Database

Data does include playoffs, so far I have 2018-19 reg season and playoffs, 2019-20 regular season
Add a column for position? Could pull this from espn

Boxscore simulater/predictor:

Create some code to look at the accuracy of predictions
Create a function to pull most recent matchup when it ends and write to matchup_results, schedule to run once a week? Or maybe it runs daily with other daily stuff, but only actually completes if the matchup just ended
Actually create a model to predict scoring based on opponent, teammates injured (at same position)
How to account for situations where more players have games than you can start on a given day
Count up games for each day, flag days where there are more than 9, and if so, drop games from the worst performing players
Don't account for ties yet
No way to get past data for testing different models (would need rosters each day, boxscores each day). But really don't need to do this, just need to find ways to better predict player performance (than just random sampling) and account for injuries/not the right amount of slots for players.

Modeling:

Factors that might influence performance: home vs away, years in the league, random intercept for opponent (some measure of within year difficulty)
Is there any temporal autocorrelation in the performance (where performance is more similar for games that are near each other in time)?

Main:

Create a "dashboard" using Rshiny
Create a trade proposer
Record DTD tag somewhere in DB and make predictions about missing games, will help with simulations. Pie in the sky use NLP to parse player updates to try and predict if they will play.

Name		Name	Last commit message	Last commit date
Latest commit History 66 Commits
.Rhistory		.Rhistory
.gitignore		.gitignore
Build database from scratch.ipynb		Build database from scratch.ipynb
ESPN FNBA API testing.ipynb		ESPN FNBA API testing.ipynb
Evaluate predictions.ipynb		Evaluate predictions.ipynb
Modeling.ipynb		Modeling.ipynb
Notes		Notes
Player rater.ipynb		Player rater.ipynb
Pull matchup results.ipynb		Pull matchup results.ipynb
README.md		README.md
Simulate_matchup_results.ipynb		Simulate_matchup_results.ipynb
Simulate_matchup_results.nbconvert.ipynb		Simulate_matchup_results.nbconvert.ipynb
Test write to db.ipynb		Test write to db.ipynb
cluj_tools.py		cluj_tools.py
octys.ipynb		octys.ipynb
player name comparison.ipynb		player name comparison.ipynb
scheduler.sh		scheduler.sh
yesterdays_boxscores.ipynb		yesterdays_boxscores.ipynb
yesterdays_boxscores.nbconvert.ipynb		yesterdays_boxscores.nbconvert.ipynb

Provide feedback